High performance Linux clusters

Meer over het boek

This book focuses on clustering for high-performance computation, with relevant insights for high-availability clustering as well. It serves as an introductory guide, providing a solid foundation in clustering concepts and practices. Key tools and best practices for exploring and expanding a system are discussed, starting with planning, hardware selection, and bulk Linux installation on multiple systems. The author highlights software that can significantly reduce deployment time, examining the pros and cons of major free software projects to aid new cluster administrators and programmers. Key projects introduced include MPI, the leading programming library for clusters, along with realistic examples and advanced usage tips. It covers OSCAR and Rocks, comprehensive installation and administrative systems, and openMosix for job distribution. The book also addresses Linux kernel extensions for load balancing, PVFS for easier clustering I/O, and C3 commands for managing multiple systems. Additional tools such as Ganglia, OpenPBS, and cloning tools (Kickstart, SIS, G4U) are included. It explores cluster installation packages in detail and provides guidelines for debugging, profiling, performance tuning, and managing jobs from multiple users, making it an immensely useful resource for anyone involved in clustering.